Supercharged Search with Typesense

TypeSense makes advanced search feel approachable and actually enjoyable, cutting through the usual confusion and frustration. You’ll dive into everything from basic queries and filters to ranking, semantic search, vector queries, and even the TypeSense Cloud—building up your skills step-by-step without feeling overwhelmed.

You install the TypeSense server locally using Homebrew, start the service, and verify it’s running with a simple curl request to the health endpoint on port 8108. Then you spin up a fresh Laravel app, set environment variables for TypeSense host, port, and API key, and pull in the official TypeSense PHP client via Composer to prep for integrating search functionality.

We grab a big books dataset from the TypeSense docs and set up a TypeSenseClient in Laravel, configuring it via config/services.php to keep environment details tidy. Then we define a books collection schema with fields like title, author, and average_rating, create that collection, and import all the book documents so we’re ready to start searching.

We organize our routes into self-contained snippets for creating, importing, and searching collections, treating them like a playground rather than production endpoints. Then we dive into searching a collection by specifying search fields, debugging import issues, and refining queries to search titles and authors. Finally, we explore sorting results by different criteria like ratings_count or publication_year, handling tie-breakers, and limiting the number of results per page for flexible, real-world search behavior.

We add search result highlighting by pulling the matched snippets from the highlight section of the response, flattening the arrays just one level, and plucking out the snippet text. Then we render the snippets as raw HTML so the matched terms get wrapped in <mark> tags, which we style with a yellow background for clear emphasis. Plus, thanks to TypeSense’s fuzzy matching, even misspelled queries still highlight relevant results nicely.

We start by building a simple search UI with a Blade view that submits a form and displays results from a TypeSense search, reading the query from the URL. The lesson walks through fetching only the book titles from the search hits, passing them to the view, and rendering them in a styled list with Tailwind via CDN—plus adding a search input that keeps its value from the query string for a smooth user experience.

We dig into adding faceted filters to the sidebar by explicitly requesting facetBy fields from our TypeSense search results, focusing on authors here. Then we map over the returned facet counts to extract filter data and pass it to the view, where we loop through and display these filters alongside the search results using a CSS grid layout. Next up, we’ll enhance this by adding checkboxes to filter results dynamically based on selected authors.

You build clickable author filters by turning each facet item into an anchor tag that merges the current query string with the selected author, ensuring the URL updates properly without losing existing search terms. On the backend, you detect the author filter in the request and adjust the TypeSense search parameters to narrow results accordingly. While this approach is simple and effective, it only supports single-author filtering, so next up is exploring a form-based method to enable multi-select filters with checkboxes.

We replace simple anchor-tag filters with a form containing checkboxes to allow selecting multiple authors for filtering search results. By dynamically generating checkbox IDs and names, submitting the form on change, and syncing the checked state using Laravel’s @checked directive, we keep the UI and query string in sync. Finally, we clean up the data structure to handle multiple authors properly and prepare for more flexible filter logic.

We set up a Vue 3 and Inertia.js project with Vite to build a live search interface that tracks user input and fetches results instantly. By binding a query ref to a search input and watching it, we trigger AJAX calls on input changes, then optimize performance with a simple debounce function to limit requests. This approach gives instant feedback without needing heavy dedicated search libraries—though those options come next.

We set up a search override that boosts a specific book to the top of results when someone searches exactly for twilight, simulating a sponsorship scenario where a publisher pays to promote their title. The override targets the book by its string ID and uses an upsert method to insert or update the override in the collection. We also cover how to remove the override later, showing how flexible this approach can be for managing search result priorities.

We set up Vue InstantSearch with the TypeSense adapter to power our search UI, pulling in widgets like SearchBox and Highlight for instant querying and result highlighting without manual debounce or request handling. Then we add a stats widget to show result counts and search speed, all seamlessly integrated and ready to extend with facets next.

Laravel Scout provides a driver-based, unified API for adding full-text search to Eloquent models, letting you swap out search engines like the database driver or TypeSense without changing your code. After installing Scout and configuring the scout.php file, you set up your model’s searchable schema, generate dummy data, and import it into TypeSense. Finally, you can perform searches easily using the searchable trait and Artisan commands like scout:import to sync your data.

We level up the basic search UI by adding Algolia’s optional instantsearch.css for quick styling and ditching manual API calls since instantsearch handles them automatically. Then, we enhance the results with highlighted book titles, authors, and publication years, and introduce faceting using the RefinementList and NumericMenu components for filtering by authors and publication year ranges—all while keeping the layout clean with Tailwind and some simple CSS tweaks. The best part? Most of the heavy lifting happens behind the scenes, letting us focus on styling and UX without writing complex search logic.

You dig into how search ranking works by starting with a text match score, then using declared default sorting like ratings_count or publication year as tiebreakers when multiple results share the same match. You also learn how to override sorting with sortBy to prioritize things like newest releases or popularity, and see why combining multiple tiebreakers—like text match, release year, then views—makes your results smarter and more relevant. Finally, you get a sneak peek at boosting or promoting specific results regardless of the algorithm, setting you up for customizing search rankings in real apps.

We set up a search override that boosts a less popular book to the top of results when someone searches exactly for twilight, simulating a paid promotion scenario. The override uses a rule targeting the query and moves the specified document ID to position one via the upsert method on the collection. Along the way, we handle key details like ensuring IDs are strings for TypeSense and show how to create, apply, and delete these overrides programmatically.

Laravel Scout offers a driver-based API to add full-text search to your Eloquent models, letting you easily swap search engines like the database driver or TypeSense without changing your code. After installing Scout and configuring the scout.php file, you create a model with dummy data, define a TypeSense schema in Scout’s config, and use the Searchable trait to sync your data. Finally, you import your records into TypeSense and perform searches right from your routes, customizing the default query field as needed.

Running a search with Scout returns an Eloquent collection of model instances, so you can work directly with your Course models instead of raw data. You can also use raw() to get the original TypeSense response if you need to dig into the hits manually. Filtering works with a simple where method to narrow results by attributes like user_id or category, and pagination feels just like Eloquent, thanks to Scout’s builder wrapping TypeSense queries.

You see how using Eloquent model events with the Searchable trait automatically syncs your database changes with a TypeSense collection—create, update, or delete a record, and Scout handles the rest behind the scenes. But if you insert or update records manually (outside Eloquent), you need to call searchable or unsearchable yourself to keep things in sync. Plus, you can control which records get indexed by overriding shouldBeSearchable on your model, letting you exclude, say, archived courses from search results.

You learn how to fine-tune your TypeSense searches by using the options method to specify fields like category or description for querying and filtering results. The lesson also shows how to customize search parameters on the fly—like changing highlight tags—and how to access raw search results for more control, all while keeping the API familiar and easy to work with.

Semantic search lets you find results that match the general vibe of a query, not just exact keywords, by converting both documents and queries into numeric embeddings. Using cosine similarity, you compare these embeddings to surface conceptually related matches, and with tools like TypeSense, you can automate embedding generation and vector search without manually handling the math. This approach helps you build smarter searches that understand context, even when the exact terms don’t appear in your data.

TypeSense Cloud takes the hassle out of managing your own search server by handling provisioning, scaling, and failover for you, plus it offers a handy dashboard for managing indices and analytics. You’ll see how to set up a cluster, generate API keys, and sync your Laravel app’s data seamlessly, then integrate with InstantSearch.js for lightning-fast, typo-tolerant search with faceting. Finally, it dives into tracking popular and zero-result queries via custom collections, giving you powerful insights to improve your search experience.